Neo4j Integration
Monitor graph query performance, page cache hit ratios, and transaction throughput across your Neo4j deployments. Get AI-powered root cause analysis for graph engine bottlenecks and Causal Cluster anomalies.
How It Works
Enable Prometheus Metrics Endpoint
Enable the Neo4j Prometheus endpoint in neo4j.conf by setting metrics.prometheus.enabled=true and metrics.prometheus.endpoint=localhost:2004. The TigerOps agent scrapes this endpoint without any additional plugins.
Deploy TigerOps Agent
Install the TigerOps agent on each Neo4j instance or as a Kubernetes DaemonSet. The agent scrapes graph engine metrics including cypher planning, cache, transactions, and Bolt connection pool stats.
Configure Causal Cluster Monitoring
For Neo4j Causal Cluster deployments, TigerOps auto-labels metrics with server role (leader/follower/read-replica) and monitors Raft replication lag and catch-up progress across all cluster members.
Set Query and Cache Alerts
Define SLOs for graph query latency, page cache hit ratio minimums, and open transaction counts. TigerOps correlates page cache evictions with query latency degradation to surface root causes instantly.
What You Get Out of the Box
Graph Query Performance
Cypher planning time, execution time, and wait time per query pattern. TigerOps tracks query plan cache hit/miss ratios and alerts when plan invalidations cause repeated expensive replanning.
Page Cache Hit Ratio
Page cache hit and miss counts, eviction rates, and flushes per second. TigerOps predicts when cache pressure will degrade query performance and recommends heap size adjustments.
Transaction Throughput
Committed, rolled-back, and concurrent transaction counts. Open transaction duration histograms detect long-running transactions that hold locks and degrade write throughput.
Bolt Connection Pool Monitoring
Active Bolt connections, idle connections, connection acquisition time, and failed handshake rates per driver pool. Alert before connection saturation causes application-level timeouts.
Causal Cluster Replication Lag
Raft log commit index lag between leader and followers, catch-up operation rates, and read-replica staleness. TigerOps fires alerts when follower lag exceeds your consistency SLO.
AI Root Cause Analysis
When graph query latency spikes, TigerOps AI examines page cache pressure, GC pause times, open transaction accumulation, and Raft leadership stability simultaneously to identify the root cause.
TigerOps Agent Config for Neo4j
Enable the Neo4j Prometheus endpoint and configure the TigerOps agent to scrape graph engine metrics.
# Step 1: Enable Prometheus metrics in neo4j.conf
# metrics.prometheus.enabled=true
# metrics.prometheus.endpoint=0.0.0.0:2004
# metrics.jvm.enabled=true
# metrics.neo4j.pagecache.enabled=true
# metrics.neo4j.tx.enabled=true
# metrics.neo4j.server.enabled=true
# Step 2: TigerOps agent config
# Place at /etc/tigerops/conf.d/neo4j.yaml
integrations:
- name: neo4j
type: prometheus_scrape
config:
targets:
- neo4j-core-0.neo4j.svc.cluster.local:2004
- neo4j-core-1.neo4j.svc.cluster.local:2004
- neo4j-core-2.neo4j.svc.cluster.local:2004
- neo4j-replica-0.neo4j.svc.cluster.local:2004
# Enrich metrics with cluster role from Bolt
cluster_enrichment:
enabled: true
bolt_uri: bolt+routing://neo4j.svc.cluster.local:7687
username: tigerops_monitor
password_env: NEO4J_MONITOR_PASSWORD
# Metric families to collect
metric_filters:
- neo4j_page_cache_*
- neo4j_transaction_*
- neo4j_bolt_*
- neo4j_cypher_*
- neo4j_causal_clustering_*
- neo4j_store_*
- jvm_*
scrape_interval: 15s
remote_write:
endpoint: https://ingest.atatus.net/api/v1/write
bearer_token: "${TIGEROPS_API_KEY}"
alerts:
pageCacheHitRatioMin: 0.95
openTransactionsMax: 100
queryLatencyP99Ms: 200
raftLagMax: 1000 # Raft log entries behindCommon Questions
Which Neo4j editions and versions does TigerOps support?
TigerOps supports Neo4j Community and Enterprise editions, versions 4.x and 5.x. Neo4j AuraDB (managed cloud) is supported via the Aura metrics API. The Prometheus metrics endpoint used by TigerOps is available in Neo4j 3.5+ (Enterprise) and 4.0+ (Community).
How does TigerOps monitor query performance without accessing query logs?
TigerOps uses the neo4j_dbms_query_execution_latency_millis histogram exposed on the Prometheus endpoint. For deeper query-level analysis, TigerOps can also query the db.listQueries() procedure via the Bolt driver on a read-only monitoring connection to get currently running query details.
Can TigerOps detect long-running transactions automatically?
Yes. TigerOps monitors neo4j_transaction_active_read and neo4j_transaction_active_write counts, and when combined with duration histograms, identifies when transactions are accumulating. Alert rules fire when the 95th percentile open transaction age exceeds your configured threshold.
How does TigerOps handle Neo4j Causal Cluster leader elections?
TigerOps tracks Raft term changes and leadership transitions as events on your metrics timeline. When a leader election occurs, it appears as an annotation on all related dashboards so you can immediately see whether a latency spike coincides with an election.
Does TigerOps support multi-database Neo4j setups?
Yes. Neo4j 4.0+ supports multiple databases per instance. TigerOps collects per-database metrics including individual store sizes, transaction counts, and query latency where Neo4j exposes them, and uses the database label for filtering in dashboards and alerts.
Get Deep Visibility Into Your Neo4j Graph Engine
Page cache analysis, Causal Cluster replication monitoring, and AI root cause analysis. Deploy in 5 minutes.